Comments for WonderProxy Blog

Comment on Miles per Milisecond, a Look at the WonderProxy Network by JD Conley

JD Conley — Mon, 27 Jun 2011 21:36:32 +0000

The other thing to take into account is that ICMP is typically downgraded in priority in routers. You may want to try a custom UDP packet based test or to-the-metal TCP implementation (so you can track retries).

Comment on Miles per Milisecond, a Look at the WonderProxy Network by Leland

Leland — Thu, 10 Feb 2011 11:39:15 +0000

The reason for the transatlantic path is because peering always takes precedence over transit. We peer with packet exchange at the LINX in London, but do not have a session with them at Ashburn. I have raised this with them to attempt to setup a session at Ashburn as well so that the north-american traffic to that destination exits via Baltimore.

Comment on Miles per Milisecond, a Look at the WonderProxy Network by Andy Delcambre

Andy Delcambre — Thu, 10 Feb 2011 06:11:31 +0000

Reminds of the story of the 500 mile email.

http://www.ibiblio.org/harris/500milemail.html

Comment on Miles per Milisecond, a Look at the WonderProxy Network by Will Roberts

Will Roberts — Thu, 10 Feb 2011 01:14:24 +0000

Tony,

These values are straight from ping without any modification, so they’ll include the time it takes to get to/from the copper as well as the time spent on the remote machine identifying the packet and responding to it. Do you have any suggestions on how to calculate that?

Our main goal was a comparison between our servers instead of absolute times. We assumed that the turnaround time at each server would be constant so it’s easy to ignore in our comparison.

Comment on Miles per Milisecond, a Look at the WonderProxy Network by Tony Li

Tony Li — Thu, 10 Feb 2011 00:54:51 +0000

This would seem to reflect propagation delay. Did you do anything to factor in transmission delays (e.g., the time it takes to actually put the packet on the fiber)? The turnaround time at the other end of the ping? I suspect that after you factor these in, you’ll see that things are very good indeed.

It would also be wonderful to understand what’s going wrong that’s causing the poor times.

Comment on Miles per Milisecond, a Look at the WonderProxy Network by Paul Reinheimer

Paul Reinheimer — Wed, 09 Feb 2011 21:20:48 +0000

Hi Steve,

Well don’t we look silly. Thanks for pointing this out, I’ve corrected the later two tables.

paul

Comment on Miles per Milisecond, a Look at the WonderProxy Network by Steve Weller

Steve Weller — Wed, 09 Feb 2011 20:25:18 +0000

Ping times are round-trip, so you must double the distances.

This makes your miles per millisecond and percentage of the speed of light tables half their correct value.

Comment on Buying VPS Systems by Xen, OpenVZ, Virtuozzo, and You! « WonderProxy Blog

Xen, OpenVZ, Virtuozzo, and You! « WonderProxy Blog — Sat, 29 Jan 2011 04:32:50 +0000

[...] we’re incredibly geographically sensitive, we don’t always get to choose our hosting providers based on their virtualization technology. When we do have the choice, we strongly prefer Xen over [...]

Comment on Squid Log Parsing for Proxy Billing by Improving Site Performance « WonderProxy Blog

Improving Site Performance « WonderProxy Blog — Thu, 20 Jan 2011 14:48:40 +0000

[...] ones that displayed aggregate statistics for all users. Looking at our usage table built on the squid logs it has nearly a million rows. Despite being indexed there’s still a lot of data to aggregate [...]

Comment on HOWTO: Managing 30+ Servers by Improving Site Performance « WonderProxy Blog

Improving Site Performance « WonderProxy Blog — Thu, 20 Jan 2011 14:48:31 +0000

[...] server setup and maintenance architecture (we’ve blogged about Setting up Proxy Servers and Managing 30+ servers previously). More recently we’ve been taking a harder look at how the site performs, both for [...]